-
Notifications
You must be signed in to change notification settings - Fork 6
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Relax minimal numpy version, if possible #40
Comments
Yeah, no need to explain yourself for using old versions of python packages, I know painfully well how disastrous the package management situation is in python :). There's no deep reason for requiring numpy ^1.19.0, it's just the oldest version of numpy I've ever tested with. I just released a new version of constriction (v0.3.3) that reduces the required numpy version to ^1.16. Please let me know if this resolves your problem (you'll need to use python 3.10 or older; for python 3.11, we require numpy ^1.23.2, and for python 3.12 we require numpy ^1.26.1; relaxing those requirements would be much more difficult). |
Perfect timing. I was about to do a fork. However, there is a problem, probably due to a possible P.S. Maybe I am barking at the wrong tree, but I cannot add |
P.P.S. It looks like you should change requirements here: Line 17 in 4077f33
See https://www.maturin.rs/metadata. |
Sorry, I should have double checked this. Thank you for hunting down the bug yourself! I fixed it in commit 3f2a96c and just released a new version (this time I did actually look into the wheel's metadata and verified it now contains "numpy~=1.16". Please let me know if this works for you. |
Great! I can confirm that I minor detail – I am unsure whether it is worth a separate issue – is the following warning from Poetry:
|
Great, thanks! I opened #43 for the issue about validating python wheels. The warning is no reason for concern, but I should indeed fix it to prevent the warning message. |
I am currently working on maintaining a legacy codebase which for valid reasons1 requires numpy (>=1.16.0,<1.19.0). I was about to introduce
constriction
as a mean to significantly reduce memory consumption by some base models, only to find out that I cannot install it due to version conflict.Is numpy>=1.19.0 a hard requirement?
Footnotes
Old versions of some scientific libraries require old version of numpy. It is impossible to upgrade those libraries because newer versions have introduced incompatible changes to serialized models. We have to support existing models and cannot recreate them. ↩
The text was updated successfully, but these errors were encountered: